• ETRI Journal
• /
• v.41 no.2
• /
• pp.235-241
• /
• 2019

In this paper, we propose a rank-weighted reconstruction feature to improve the robustness of a feed-forward deep neural network (FFDNN)-based acoustic model. In the FFDNN-based acoustic model, an input feature is constructed by vectorizing a submatrix that is created by slicing the feature vectors of frames within a context window. In this type of feature construction, the appropriate context window size is important because it determines the amount of trivial or discriminative information, such as redundancy, or temporal context of the input features. However, we ascertained whether a single parameter is sufficiently able to control the quantity of information. Therefore, we investigated the input feature construction from the perspectives of rank and nullity, and proposed a rank-weighted reconstruction feature herein, that allows for the retention of speech information components and the reduction in trivial components. The proposed method was evaluated in the TIMIT phone recognition and Wall Street Journal (WSJ) domains. The proposed method reduced the phone error rate of the TIMIT domain from 18.4% to 18.0%, and the word error rate of the WSJ domain from 4.70% to 4.43%.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2014.10a
• /
• pp.569-569
• /
• 2014

The interior vehicle noise due to the exterior aerodynamic field is an important topic in the acoustic design of a car. The air flow detached from the A-pillar and impacting the side windows are of particular interest as they are located close to the driver / passenger and provides a lower insulation index than the trimmed car body parts. This paper presents a numerical analysis method for a simplified vehicle model. The internal air cavity including trim component are included in the simulation. The car body includes the windshield and two side windows. The body is made of aluminum and trimmed with porous layers. The methodology proposed in this paper relies on two steps: the first step involves the computation of the exterior flow and turbulence induced non-linear acoustic field using CFD Code. The second step consists in the computation of the vibro-acoustic transmission through the window using the finite element vibro-acoustic solver Actran.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.26 no.7
• /
• pp.765-773
• /
• 2016

The vehicle interior noise caused by exterior fluid flow field is one of critical issues for product developers in a design stage. Especially, turbulence and vortex flow around A-pillar and side mirror affect vehicle interior noise through a side window. The reliable numerical prediction of the noise in a vehicle cabin due to exterior flow requires distinguishing between the aerodynamic (incompressible) and the acoustic (compressible) surface pressures as well as accurate computation of surface pressure due to this flow, since the transmission characteristics of incompressible and compressible pressure waves are quite different from each other. In this paper, effective signal processing technique is proposed to separate them. First, the exterior flow field is computed by applying computational aeroacoustics techniques based on the Lattice Boltzmann method. Then, the wavenumber-frequency analysis is performed for the time-space pressure signals in order to characterize pressure fluctuations on the surface of a vehicle side window. The wavenumber-frequency diagrams of the power spectral density shows clearly two distinct regions corresponding to the hydrodynamic and the acoustic components of the surface pressure fluctuations. Lastly, decomposition of surface pressure fluctuation into incompressible and compressible ones is successfully accomplished by taking the inverse Fourier transform on the wavenumber-frequency diagrams.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2012.04a
• /
• pp.763-768
• /
• 2012

Active window system which can reduce the environmental noises, such as traffic noise and construction noise, from an open window into a room was proposed in the previous works. The key idea of the proposed active window system was that the control sources are approximately collocated with the primary noise source in terms of the acoustic power for global noise reduction throughout the interior room. Moreover, because it is important not to intrude into the living space in the building environment, no error sensors were used and an open-loop control method using control sources at the window frame and the reference sensors outside the room was used for the proposed system. The open-loop control gain was calculated by the interior room model assumed as the semi-infinite space, and the interior sound field was estimated by Rayleigh integral equation under the baffled window model assumption. However, windows with a finite thickness should were considered for the calculation of the open-loop control gain of the active window system since these are representative of most window cases. Therefore, the finite thickness window model based on the Sgard's model was derived and the open-loop control gain using the interior sound field estimated by that model was calculated for active window system. To compare the performance of these two models, a scale-model experiment was performed in an anechoic chamber according to noise source directions. Experimental results showed that the performance for the thickness window model is better than the baffled window model as the angle with respect to the perpendicular direction is larger.

• The Journal of Engineering Geology
• /
• v.25 no.1
• /
• pp.103-113
• /
• 2015

This study presents a novel algorithm for determining the P-wave arrival time using amoving window function to improve source localization in low-SNR (signal-to-noise ratio)acoustic emissions. The proposed algorithm was applied to low-SNR signals to verify the accuracy of measurements against existing algorithms. When other algorithms were applied, the test results revealed that SNR decreased and accuracy was reduced, especially where SNR wasless than 2.14. The proposed algorithm using amoving window function considers the frequency characteristic and signal amplitude simultaneously, and produced reliable results where SNR was 2.14.

• International Journal of Naval Architecture and Ocean Engineering
• /
• v.13 no.1
• /
• pp.187-193
• /
• 2021

In order to solve the problem of false terrains caused by environmental interferences and tunneling effect in the conventional multi-beam seafloor terrain detection, this paper proposed a seafloor topography detection method based on fast two-dimensional (2D) Censored Mean Level Detector-statistics Constant False Alarm Rate (CMLD-CFAR) method. The proposed method uses s cross-sliding window. The target occlusion phenomenon that occurs in multi-target environments can be eliminated by censoring some of the large cells of the reference cells, while the remaining reference cells are used to calculate the local threshold. The conventional 2D CMLD-CFAR methods need to estimate the background clutter power level for every pixel, thus increasing the computational burden significantly. In order to overcome this limitation, the proposed method uses a fast algorithm to select the Regions of Interest (ROI) based on a global threshold, while the rest pixels are distinguished as clutter directly. The proposed method is verified by experiments with real multi-beam data. The results show that the proposed method can effectively solve the problem of false terrain in a multi-beam terrain survey and achieve a high detection accuracy.

• The Journal of the Acoustical Society of Korea
• /
• v.26 no.1
• /
• pp.8-15
• /
• 2007

Speech intelligibility of the eavesdropping sound is investigated on a acoustic cavity - glass window coupled system. Using MLS (Maximum Length Sequency) signal as a sound source, acceleration and velocity responses of the glass window are measured by accelerometer and laser doppler vibrometer. MTF (Modulation Transfer Function) is used to identify tile speech transmission characteristics of the cavity and window system. STI (Speech Transmission Index) based upon MTF is calculated and speech intelligibility of the vibration sound of the glass window is estimated. Speech intelligibilities by the acceleration signal and the velocity signal are compared. Finally, intelligibility of the conversation sound is confirmed by the subjective test.

• Transactions of the Korean Society for Noise and Vibration Engineering
• /
• v.21 no.9
• /
• pp.820-827
• /
• 2011

Recently, noise has been regarded as one of the most notorious and frequent environmental pollutions which can be often encountered not only in the living space but also in the industrial site. Studies on physiological and psychological effects of long-term noise exposure to human being have commanded the public interest on noise issues. Since environmental noises such as traffic noise and construction noise is mainly flowed through the open window, it is necessary to develop the active noise control system to reduce it inside the building. Although control speakers and microphones for the noise signal measurement in the control region are essential for the conventional active noise control methods, it is impossible to implement them in the control region in the building environment because the control region is the living quarter and they may hinder activities of the residents. Therefore, we proposed the active window system to reduce the exterior noise flowed through the open window with microphones installed outside the window and control speakers installed at the frame of the window. To confirm the performance of the proposed active window, we carried out the simulation and experiment using active window system with 8 control speakers. Simulation results showed the noticeable noise reduction effect inside the control region within the frequency range without the spatial aliasing. Experimental result showed that the total acoustic potential energy inside the room of the scale model is reduced to about 10dB within the interest of frequency range.

• Proceedings of the Korean Society for Noise and Vibration Engineering Conference
• /
• 2014.10a
• /
• pp.670-672
• /
• 2014

In this study, we investigate active control of noise transmitted through a window of enclosures minimizing the acoustic power. To reduce noise of the enclosures, passive methods with absorbing material are generally used. The passive methods, however, are limited use due to the vantilation windows. In this case, these windows are path of noise leakage. Feedforward active noise control technology is applied to minimize the sound power from the enclosure. The feedforward controller is implemented with FIR filter based on the transfer functions calculated numerically. The controller reflects the delay due to FIR filter. The noise transmitted through the window is actively controlled, and the reduction of the power is obtained by 15dB.

• The Journal of the Acoustical Society of Korea
• /
• v.21 no.8
• /
• pp.744-756
• /
• 2002

The transient acoustic fields formed by a 3.5 ㎒ curved linear array transducer which is commonly used in ultrasonic medical imaging system for diagnosis of abdomen are systematically analyzed to obtain new design parameters for the better acoustic image. In the analysis with an assumption of radiating waveform, element size, radius of curvature, amplitude apodization are considered as parameters giving constitutive relations with the fields. As simulation results, appropriate new parameters with the reduced curvature and elevation aperture and the apodization of Hamming window, which make an improved acoustic beam with lower side lobe levels than a conventional typical transducer, are obtained.